Le dimanche 29 janvier 2012 à 06:11 +0100, Wolfgang Bornath a écrit : > 2012/1/28 Michael Scherer <[email protected]>:
> > While I appreciate Wobo efforts to get meaningful numbers, I am rather > > doubtful about a few things. First, the choice of questions make me > > wonder, since my memories tell me that for doing such kind of polls, you > > first need to do preliminary research to find the options. > > The questions are a result of reading user's postings for quite some years. So basically, that's "because I know that's this, based on what I remember". But the formal methodology is a little bit more complex than that, mainly to avoid any specific bias due to the selection by the peson doing the poll. For example, the diversity of the group may be important, as well as the cultural norms. If you focus just on forums, you focus just on a certain type of users. Having numbers to present is also usually important to show your result as meaningful. > > There is for example various things like : > > "networking out of the box", and "good hardware recognition" that are > > the same. > > Not at all. Example: my wifi hardware is recognized out of the box but > it is a drag to get it working. Then why treat networking differently than any others functions like printing ? And what make you think that your specific issue is not linked to hardware problem ? Good hardware goes beyond "the interface appear in ifconfig, therefore, it is supported perfectly". That's usually the contrary, the whole drakxnet/wpa_supplicant/whatever bit work for some people and not from others, and the only stuff that change is the wifi chipset. That's likely a hardware support issue. Main developer of Network Manager explained that on his blog. People often tend to think that nm is broken, while the problem is usually a bug in the driver that is triggered by nm. Hence, the question do not depict exactly what should be improved, just what some people think the problem is, without giving more informations on exactly the problem is. The poll is not the right tool for that, the bugzilla is ( and we are fully aware of the bugs that are reported ). > > And LTS go hand to hand with "stable as top priority" ( since LTS mean > > the system do not change for a long time, and usually, once enough bugs > > are fixed, softwares are rather stable ). > > Not the same. One is a consequence of the other, in my vision of a LTS ( and that's also the vision of Canonical, which is basically the only definition we have of LTS so far ). And as seen on Mandriva list or on the forum, LTS mean nothing for some users, and mean different things for different people ( ie, long term, how much is this ? support, what do it entail exactly ? ). For people that see LTS as "I never reinstall and get new versions of some stuff", there is a poll item. For those that say "I never reinstall and system do not move", there is a poll item. So LTS can ( and rightfully ) be seen as a technical detail to achieve what people want, hence the few vote ( which doesn't reflect what people asked in others time ). > > Not to mention that a poll may give incorrect expectation to people, and > > usually only measure a fraction of the users. > > Yes, same as elections and marketing surveys. Except for elections, you know who you ask and how many people, and that's to select someone, based on a large corpus of proposition, that's rather different. So that's more like a marketing survey, without enough detailed information ( like "we asked to people who are this age, this gender, this level of skills ). We will know what some people want, but not who they are. And without knowing who they are, we will not know if they are the ones we want, if the survey results are biased, or what kind of bias we should be careful of. For example, do we want to listen to people that are here since the start, or not ? To people that are outside of the current community, or those that are unsatisfied with their current distribution ? > > And more importantly, it doesn't help us on the crucial thing ( IMHO ), > > how to expend the contributeurs pool. > > I think asking "where would you want to put efforts in the next year for > > Mageia" would allow us to better see what kind of priority we should > > have to grow the community. ( followed by discussions of "why don't you > > do it now" that would help finding potential weak spots in the system ). > > > > This or asking directly, "why don't you work on $FOO". Users will come > > when we have a great product ( ie, grow naturally up to a point ). And > > product will be great with enough contribution, hence the focus on that. > > This was never the goal of this poll. I know this is not. However, that's a poll that would be more useful, since again, our first and foremost goal is to have contributions. Without this, the rest will not be > Why don't you start such a poll yourself? You mean "why don't I start a vast social engineering campaign to recruit contributers ? " I already do it ( but I guess people do not see it ), but I never go to forums. I would not like people to step on my toes on sysadmin side, so I have the politeness of letting those that run the forums to take care of that. > @ALL: Thx for your opinions though I wish you'd offer these there > where the poll is and not where I merely gave a pointer to the poll. Personally, I will continue here because the lack of thread in forum will make hard to have a meaningful discussion. I answered to the poll, and to the forum, but I do not think I can be as precise ( or read "verbose") there than I am on the ml. -- Michael Scherer
